|
Activity Number:
|
136
|
|
Type:
|
Invited
|
|
Date/Time:
|
Monday, August 3, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #303200 |
|
Title:
|
The Infinite Dimension of the Data Revealed by a Perfectly Executed Measurement of a Scalar Variable
|
|
Author(s):
|
Constantine Frangakis*+
|
|
Companies:
|
Johns Hopkins University
|
|
Address:
|
Department of Biostatistics, Baltmore, MD, 21205,
|
|
Keywords:
|
|
|
Abstract:
|
We consider a study with a known, planned measurement mechanism to record a "planned variable" of finite dimension. Current theory faced with missing data implicitly assumes that we can completely describe the available data from such a study using an "extended variable" including both the planned variable, and the indicator of success or failure to measure the planned variable. We show here such an extension will be always internally incomplete. This means that there will always be an unplanned variable that is revealed simply by the previous extension -this variable is the success or failure to measure the previously "extended variables." As a result, we show that there is an infinite dimension of data that is revealed simply from the plan to measure even a scalar variable, and even if the measurement plan has worked perfectly. We discuss the implications on inference through example.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2009 program |
